ISL1 Is Necessary for Maximal Thyrotrope Response to Hypothyroidism

Abstract

ISLET1 is a homeodomain transcription factor necessary for development of the pituitary, retina, motor neurons, heart, and pancreas. Isl1-deficient mice (Isl1^−/−) die early during embryogenesis at embryonic day 10.5 due to heart defects, and at that time, they have an undersized pituitary primordium. ISL1 is expressed in differentiating pituitary cells in early embryogenesis. Here, we report the cell-specific expression of ISL1 and assessment of its role in gonadotropes and thyrotropes. Isl1 expression is elevated in pituitaries of Cga^−/− mice, a model of hypothyroidism with thyrotrope hypertrophy and hyperplasia. Thyrotrope-specific disruption of Isl1 with Tshb-cre is permissive for normal serum TSH, but T₄ levels are decreased, suggesting decreased thyrotrope function. Inducing hypothyroidism in normal mice causes a reduction in T₄ levels and dramatically elevated TSH response, but mice with thyrotrope-specific disruption of Isl1 have a blunted TSH response. In contrast, deletion of Isl1 in gonadotropes with an Lhb-cre transgene has no obvious effect on gonadotrope function or fertility. These results show that ISL1 is necessary for maximal thyrotrope response to hypothyroidism, in addition to its role in development of Rathke's pouch.

The pituitary gland is a master organ that controls the endocrine organs of the body. The anterior lobe secretes 6 major hormones that regulate various processes including growth, lactation, stress response and reproduction, whereas the posterior lobe regulates mainly water homeostasis (1). Pituitary organogenesis is a complex process requiring temporal and spatial regulation of several signaling pathways and transcription factors. Four LIM homeodomain transcription factors are required for normal pituitary development: LHX2, LHX3, LHX4, and ISL1. The LIM name comes from the names of the proteins in which the zinc finger protein interaction domains were first discovered: Lin11, Isl1, and Mec3. LHX2 has a role in pituitary posterior lobe development (2). Lhx3 and Lhx4 are crucial in the early steps of pituitary ontogenesis and have dosage sensitive, partially overlapping functions in development of the anterior pituitary primordium called Rathke's pouch (3–6). Loss of Lhx3 or Lhx4 expression in mice results in pituitary hypoplasia, increased cell death, cell differentiation defects, and ectopic ISL1 expression (4–7). Humans with loss-of-function LHX3 or LHX4 mutations have similar pituitary phenotypes to the mouse models. Patients homozygous for LHX3 mutations have variable pituitary defects that include hypopituitarism, combined pituitary hormone deficiency, and a hypoplastic anterior pituitary gland. Patients can also exhibit other phenotypes that include hearing loss and reduced neck rotation (8–11). Patients heterozygous for LHX4 mutations also have variable pituitary phenotypes that include combined pituitary hormone deficiency and a hypoplastic anterior pituitary gland, and in some cases, ectopic posterior pituitary gland (12–15).

Isl1 was identified more than 20 years ago in a cDNA library derived from insulin producing cells (16). Since then, roles for ISL1 have been defined in the development, proliferation, function and maintenance of specialized cells in the pancreas, retina, inner ear, motor neurons, and heart (17, 18). Isl1 is required for pancreatic ISLET cell differentiation and is highly expressed in mesenchymal stem cells of human pancreas (19–21). Isl1 expression in the eye is necessary for retinal ganglion cell differentiation (22, 23). Cardiovascular progenitors expressing ISL1 give rise to the cardiomyocyte, pacemaker, smooth muscle and endothelial cell lineages (24, 25). ISL1 is indispensable for the craniocervical and spinal motor neuron formation in the neural tube and for interneuron differentiation (18). ISL1 is also expressed in various regions of the hypothalamus. It is necessary for melanocortin neuron cell fate and for estrogen receptor-α expression (26, 27). In all these organs, ISL1 is expressed in the early steps of organogenesis, and it is crucial for the differentiation process leading to a mature and functional organ.

Isl1 is expressed early in pituitary development and is required for development of Rathke's pouch (28). Isl1 mutants have a small, rudimentary pouch, and the affected embryos die at e10.5 due to heart defects. Thus, the role of Isl1 has not been assessed at later stages of pituitary growth and development, and the cell specificity of expression has not been fully analyzed. In mice, Isl1 is expressed in the oral ectoderm that gives rise to the pouch at embryonic day (e)9.5 and then is expressed in differentiating cells (29). A great portion of these cells express the α-subunit of the pituitary glycoprotein hormones (officially known as chorionic gonadotropin-α [CGA]) and give rise to POU1F1-independent rostral tip thyrotropes (30). In sheep and chick, ISL1 is expressed in thyrotropes and gonadotropes (31, 32), which suggests a role for ISL1 in these cell types. Gene expression analysis in immortalized cell lines revealed that ISL1 is enriched in gonadotrope-like cells and can regulate gonadotropin releasing hormone receptor expression, but it is not detected in corticotrope or somatomammotrope-like cells (33).

Here, we report a detailed analysis of Isl1 expression in pituitaries of normal and hypothyroid mice and test the requirement for Isl1 in pituitary cell maintenance and function. We detect ISL1 in thyrotropes and gonadotropes during development. We discovered that levels of ISL1 transcripts are substantially elevated under conditions of thyrotrope hypertrophy and hyperplasia. Inactivation of Isl1 in thyrotropes causes mild hypothyroidism, reduced pituitary TSH response to hypothyroidism, and reduced growth. In contrast, mice with ISL1-deficient gonadotropes have normal fertility. In addition to its established role in early pituitary organogenesis, these results suggest that ISL1 has a role in thyrotrope function after birth.

Materials and Methods

Mice

All mice were maintained at the University of Michigan under the guidelines of the Unit for Laboratory Animal Medicine and the University Committee for Care and Use of Animals. Tg(Pou1f1-cre) mice were donated by Paul LeTissier (34). Tg(Lhb-cre) mice were previously described in Charles et al (35). Tg(Tshb-cre) mice were previously described in Castinetti et al (36). The Isl1^loxp/loxp were donated by Lin Gan (37). The Isl1^+/− mice were generated by mating the Isl1^loxp/loxp mice to B6.C-Tg(CMV-cre)1Cgn/J mice (stock number 006054; The Jackson Laboratory). The B6.129S4-Gt(ROSA)26Sor^tm1Sor/J mice (R26R/R26R) were purchased from The Jackson Laboratory (stock number 003474) and maintained as homozygotes.

Tissue preparation and histology

All tissues were fixed in 4% paraformaldehyde in 1× PBS. Pituitaries and thyroids were collected at 4 and 8 weeks of age and fixed for 1 hour. Embryos collected at e11.5 were fixed for 30 minutes, and later stages, e12.5, e13.5, and e14.5, were fixed for 1 hour, and e16.5 embryos were fixed for 2 hours. Heads collected from postnatal day (P)1, P2, and P7 neonates were fixed for 4 hours. Tissues were rinsed in 1× PBS, dehydrated with series of graduated ethanols, and embedded in paraffin. The frozen tissues were rinsed in 1× PBS, then embedded in Optimum Cutting Temperature Compound (Tissue-Tek), frozen on dry ice, and stored at −80°C. Adult pituitaries and heads from neonates were sectioned in the coronal plane at 5-μm thickness for paraffin and at 16-μm thickness for frozen sections. Embryos collected at e11.5, e12.5, e13.5, e14.5, e16.5, and adult thyroid glands were sectioned in the sagittal plane at 5-μm thickness for paraffin and at 16-μm thickness for frozen.

At least 3 animals of each age and genotype were analyzed by immunostaining and hematoxylin and eosin (H&E) staining. The most midsagittal and midcoronal sections were used from embryos and pituitaries, respectively. The sections presenting the greatest area of thyroid tissue were used for quantifying follicular area. Slides used for immunohistochemistry were dewaxed in xylene and then rehydrated through 100% and 95% ethanols before washing in PBS. Please see Table 1 for specifics for each experiment, including antigen retrieval, endogenous peroxidase inactivation, blocking, antibody source and concentration, and detection methods. After immunohistochemistry, 4',6-diamidino-2-phenylindole counterstaining was performed for 5 minutes. After 3 washes in PBS, slides were mounted with fluorescent mounting media, and images were captured using a Leica DMRB fluorescent microscope.

Table 1.

Immunohistochemistry Experiments

Experiment	Citric Acid Boil	CH3OH: H2O2	Block	Primary Antibody	Secondary Antibody and Detection	Biotin Block	Other Primary Antibody	Other Secondary and Detection
ISL1	10 min-embryo 15 min-postnatal	Yes	M.O.M.,a TNBb	1:600 anti-ISL1, DSHBc	Antirat biotin, TSA FITC
ISL1 and POU1F1g	10 min-embryo 15 min-postnatal	Yes	M.O.M., TNB	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC	Yes	1:500 anti-POU1F1, gift from S. Rhodes	Antirabbit biotin, TSA TRITC
ISL1 and NR5A1g	10 min	Yes	M.O.M., TNB	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC	Yes	1:1000 anti-NR5A, gift from G. Hammer	Antirabbit biotin, TSA TRITC
ISL1 and TBX19g	10 min	Yes	M.O.M., TNB	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC	Yes	1:200 anti-TBX19, gift from J. Drouin	Antirabbit biotin, TSA TRITC
CGA and ISL1	10 min	Yesf	M.O.M., TNB	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC	Yes	1:1000, anti-CGA, NHPPd	Antirabbit biotin, TSA TRITC
TSHb and ISL1	15 mine	Yesf	M.O.M., TNB	1:1000 anti-TSHb, NHPP	Antirabbit biotin, TSA TRITC	Yes	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC
LHb and ISL1	15 mine	Yesf	M.O.M., TNB	1:500 anti-LHb, NHPP	Antiguinea pig, TSA TRITC	Yes	1:600 anti-ISL1, DSHB	Antirat biotin, TSA FITC
GH	none	Yes	NGSh	1:1000 anti-GH,j NHPP	Antihuman biotin, SA Cy2i
TSH	none	Yes	NGS	1:1000, anti-TSH, NHPP	Antirabbit biotin, SA Cy2i
TSH and POU1F1	10 mine	Yesf	NGS M.O.M.	1:1000, anti-TSH, NHPP	Antiguinea pig, TSA TRITC	Yes	1:500, anti-POU1F1	Antirabbit biotin, TSA FITC
TSH and NR5A1	10 mine	Yesf	NGS M.O.M.	1:1000, anti-TSH, NHPP	Antiguinea pig, TSA TRITC	Yes	1:1000, anti-NR5A1	Antirabbit biotin, TSA FITC

^aMouse kit from Vector Labs.

^bBlock from PerkinElmer Tyramide Signal Amplification TSA Fluorescein isothiocyanate FITC and Tetramethylrhodamine (TRITC) kit.

^cDevelopmental Studies Hybridoma Bank.

^dNational Hormone and Pituitary Program.

^eCitrate boiling done after primary antibody.

^fSecond CH₂OH:H₂O₂ block done before second primary antibody.

^gPrimary antibodies incubated at same time.

^h5% normal goat serum, 3% BSA in 1× PBS.

ⁱStreptavidin cyanine 2.

^jAnti-GH antibody left on for 1 hour.

For H&E (Sigma-Aldrich) staining the paraffin embedded tissue sections were dewaxed, and rehydrated, soaked in hematoxylin for 3 minutes, rinsed in distilled water, “blued” using a combination of Differentiation Solution (Sigma-Aldrich) and Scott's Tap Water (Sigma-Aldrich) with a distilled water rinse in between, soaked in eosin for 20 seconds, and rinsed in distilled water. The slides were gradually dehydrated in a series of ethanols of increasing concentration, soaked in xylene and mounted with xylene:permount 1:2 (Fisher) media. X-gal (5 bromo-4-chloro-3-indolyl-b-D-galactopyranoside) staining and neutral red staining were performed on frozen tissue as previously described (35, 38).

Thyroid follicle measurement

Adult thyroid glands were collected at 4 and 8 weeks of age. H&E-stained thyroid sections were analyzed using the ImageJ software (National Center for Biotechnology Information). The area of individual follicles was measured in 3 sections taken from the middle of the gland that were 5 sections apart (n = 3 for control, n = 4 for Isl1^loxP/loxP;Tg(Tsh-cre)). A range of 627–729 follicles was counted in 3 Isl1^loxp/loxp controls, and a range of 466–895 follicles was counted in 4 Isl1^loxP/loxP;Tg(Tsh-cre) mutants. The average follicle area was determined by dividing the total follicle area by the number of follicles measured. The Mann-Whitney U test was used for analysis. P value was set to .05.

Real-time PCR

8-week old pituitaries and livers were dissected and stored in RNALater at −20°C. Four pituitaries and 4 livers from Isl1^loxP/loxP;Tg(Tshb-cre) and Isl1^+/loxP or Isl1^loxP/loxP mice were used. RNA was isolated and deoxyribonuclease I treated using the protocols provided in the RNAqueous 4 PCR kit (Life Technologies). The cDNA was generated using the Superscript II reverse transcription system from Invitrogen (Life Technologies). Specifically, RNA and oligo dT primers were denatured at 70°C for 10 minutes and then placed on ice; 100mM dithiothreitol, 10mM deoxynucleotide triphosphates 5× first strand buffer, and 1-U Superscript II were added to the RNA primer mix. Samples were incubated at 42°C for 50 minutes, then heat inactivated for 10 minutes at 70°C. Samples without reverse transcriptase were included as negative controls. The quality of the cDNA reaction and effectiveness of the DNase I treatment of the RNA was tested using PCR with primers to the housekeeping gene, hypoxanthine guanine phosphoribosyl transferase. Quantitative PCR (qPCR) was carried out using TaqMan gene expression assays for Igf1 (Mm00439559_m1), Gh (Mm00433590_g1), Tshb (Mm00437190_m1), and Gapdh (4308316). The qPCR was set up in triplicate with 50-ng cDNA per reaction using TaqMan Universal PCR Master Mix and default run parameters on the ABI 7500 Real-Time PCR instrument. Results were expressed as fold changes compared with Gapdh using the delta cycle threshold method as described earlier (22). Triplicates containing the Master Mix and water with the TaqMan primer sets were used as a negative controls. The fold change of expression in Isl1^loxP/loxP;Tg(Tshb-cre) (n = 4) compared with Isl1^+/loxP and Isl1^loxP/loxP (n = 4) mice was calculated as described previously (39). All reagents and instruments were obtained from Life Technologies.

Cell counts

Immunopositive cells from of 3–4 sections spanning the pituitary from Isl1^+/loxP;Tg(Tshb-cre) (n = 4) and Isl1^loxP/loxP;Tg(Tshb-cre) (n = 3) mice were quantitated using ImageJ software (Leica). The number of cells coimmunostained for TSH and ISL1 were compared with the total number of cells immunopositive for TSH. The average percentage of double immunopositive cells is displayed. Statistical analysis was performed using the Student's t test, and significance level was set to 0.05.

Serum hormone measurements

After euthanasia with CO₂, blood was collected by cardiac puncture while the heart was still beating. After clotting, the blood was stored at 4°C for 24 hours and was centrifuged at 8000g for 10 minutes. The serum was stored at −20°C until use. A total of 5 μL of serum was used to analyze the total T₄ concentration (MP Biomedicals), and 20 μL of serum were used to analyze the TSH concentration (Millipore). The T₄ measure was performed in triplicates, whereas the TSH level was a single measurement.

Hypothyroidism challenge

Low iodine diet enriched in propylthiouracil (PTU) (0.15%) (Harlan Lab) was given to 3- to 8-week-old Isl1^loxP/loxP;Tg (Tshb-cre) and control mice for 4 weeks (40). Blood collection and hormone measurements were performed as described above.

Statistical analysis of hormone measurements and body weights

Data are presented as mean ± SD. Student's t test, ANOVA, or Mann-Whitney U test were used for statistical analysis where appropriate. Data were analyzed with SPSS version 17.0. ImageJ software, and Excel version 14.3.9. P < .05 was accepted as significant.

Results

ISL1 is expressed during pituitary development

We evaluated the expression of ISL1 in the mouse pituitary during development and at early postnatal age by ISL1 immunostaining (Figure 1). From e11.5 to e13.5, ISL1 immunoreactivity is detected in Rathke's pouch among the cells in the differentiation zone but not cells in the proliferative zone (41). At later stages, e14.5, e16.5, and P3, there are scattered ISL1-expressing cells in the anterior lobe. There are traces of ISL1 immunostaining in the intermediate lobe at P3, and none in the posterior lobe. The ventral hypothalamus, located rostral to the pituitary gland, exhibits robust ISL1 staining throughout development.

Figure 1.

ISL1 is expressed in Rathke's pouch during development. ISL1 immunostaining (green) and DAPI counterstaining (blue) was carried out in sections from developing pituitary glands collected from e11.5 to P3. The photographs of sagittal e11.5, e12.5, and e13.5 sections were taken at ×200, and the later ages, sagittal e14.5, e16.5, and coronal P3, at ×100. Scale bars, 100 μm. White box inset of P3 photo was taken at ×630. RP, Rathke's pouch; VD, ventral diencephalon; AL, anterior lobe; IL, intermediate lobe; PL, posterior lobe; VH, ventral hypothalamus.

To determine which pituitary cell types express ISL1, we carried out immunohistochemistry for ISL and lineage-specific transcription factors at e16.5 and evaluated nuclear colocalization (Figure 2, A–H). ISL1 colocalized with a few POU1F1⁺ cells (Figure 2, A and B), but most of the immunopositive cells expressed either ISL1 or POU1F1 but not both. ISL1 immunostaining did not colocalize with any of the TBX19⁺ cells, which mark future melanotropes and corticotropes (Figure 2, C and D) (42). ISL1 colocalized with about half of the NR5A1⁺ cells, which differentiate into gonadotropes (43), at e16.5 (Figure 2, E and F), but less colocalization was observed at P7 (Figure 2I). ISL1 colocalized with many of the CGA⁺ cells at e16.5 (Figure 2, G and H) and with many TSHb cells at P7 (Figure 2J). These data suggest that ISL1 might be involved in thyrotrope function and in the initiation of the gonadotrope differentiation but not maintenance.

Figure 2.

ISL1 is expressed in early thyrotropes and gonadotropes. A–H, ISL1 immunostaining (green) was carried out together with lineage-specific transcription factor staining (red) and the common subunit to thyrotropes and gonadotropes (CGA) on sections from e16.5 embryos. A and B, Coimmunostaining of ISL1 (green) and POU1F1 (red) is represented by the yellow nuclei (left panels at low [×100] and right high [×630] magnification). C and D, No colocalization was detected for nuclear ISL1 and TBX19. E–H, Colocalization was detected for ISL1 and NR5A1 (yellow nuclei) as well as for ISL1 (nuclear, green) and CGA (cytoplasmic, red). I, The bottom left panel shows limited coimmunostaining of ISL1 and NR5A2 postnatally (P1). J, The bottom right panel shows coimmunostaining of ISL1 and TSH postnatally (P7). White arrows point to colocalized cells. Both bottom panel images were taken at ×630. Scale bars for ×100 images are 100 μm and for ×630 images are 50 μm.

To explore the role of ISL1 in thyrotropes, we quantified Isl1 transcripts in Cga^−/− mice (44). These mutants are hypothyroid and hypogonadal due to the inability to secrete bioactive glycoprotein hormones and have an ancillary deficiency in GH and PRL (44). The pituitary glands of these mice undergo profound thyrotrope hypertrophy and hyperplasia, and have very few somatotropes and lactotropes (38, 45–47). Pituitaries were collected from 8-week-old mutant and wild-type mice (3 males and 3 females of each genotype, 12 samples total) and analyzed for Isl1 expression by qPCR using TaqMan assay. Isl1 transcripts were elevated 2.6-fold (SEM, 0.7; P = .001) in mutant pituitaries, suggesting a role in thyrotrope hypertrophy and hyperplasia.

To understand the specific roles of ISL1 in the anterior pituitary gland, we inactivated Isl1 using a Pou1f1-specific cre recombinase transgenic line (Tg[Pou1f1-cre]) (34), a gonadotrope-specific cre line (Tg[Lhb-cre]) (35), and a thyrotrope-specific cre line (Tg[Tshb-cre]) (36).

Deletion of Isl1 using Pouf1-cre had no effect on pituitary development

To characterize the developmental onset and pituitary specificity of cre activity, we crossed mice carrying cre recombinase under the control of the Pou1f1 promoter, Tg(Pou1f1-cre) (34), with a cre reporter strain consisting of a floxed stop sequence, which upon cre mediated excision permits lacZ expression (R26R/R26R) (48). We assessed the efficiency and penetrance of cre-mediated recombination by X-gal staining of frozen sections (Supplemental Figure 1A). Pou1f1-cre activity is first evident at e14.5 in the developing anterior lobe as indicated by the blue X-gal staining. Not many of the cells appear X-gal positive at e14.5. At 4 weeks, the X-gal staining appears throughout the anterior lobe, indicating cre activity beyond the cell types expected for the Pou1f1 lineage, consistent with previous reports (49). The Pou1f1-cre also exhibits unexpected activity in the thyroid gland in adult mice.

Mice heterozygous for Isl1 (Isl1^+/−) were mated with Tg(Pou1f1-cre) to produce Isl1^+/−;Tg(Pou1f1-cre) mice. We crossed Isl1^+/−;Tg(Pou1f1-cre) mice with mice homozygous for a floxed Isl1 allele (Isl1^loxP/loxP) to generate the conditional knockout Isl1^−/loxp;Tg(Pou1f1-cre) mice which are designated as mutant. Isl1^−/loxp, Isl1^+/loxp, and Isl1^+/loxp;Tg(Pou1f1-cre) were used as control mice. We analyzed the pituitaries of neonates to confirm that ISL1 was effectively ablated in the POU1F1 cells (Supplemental Figure 1B). At P2, about one third of the ISL1⁺ cells normally express POU1F1 (yellow arrows), and there are many POU1F1 cells that do not express ISL1 (pink arrow). No cells coexpressing ISL1 and POU1F1 were detected in the P2 pituitaries of Isl1^−/loxP;Tg(Pou1f1-cre) mice, indicating efficient ablation of ISL1. As expected, ISL1⁺, POU1F1⁻ cells are present (green arrow). No difference was detected in TSH immunostaining in the pituitary gland at P1 between the control and mutant groups (n = 3 for each genotype) (data not shown). The weight of control and mutant mice were the same at 5 and 8 weeks of age, suggesting that the pituitary production of TSH and GH was sufficient to support normal growth (Supplemental Figure 1C). As expected, males of both genotypes were larger than females, suggesting that sexually dimorphic regulation of GH secretion was intact. We used immunohistochemistry to quantify the number of POU1F1 cells that are TSH⁺. Surprisingly, only about half of the TSH cells were also POU1F1⁺ (Supplemental Figure 1D, white arrow). Some of the TSH⁺/POU1F1⁻ cells are NR5A1⁺, but only a few cells expressed both proteins (Supplemental Figure 1D). These results suggest that thyrotropes are heterogeneous and that effective deletion of ISL1 in thyrotropes could be better achieved with a Tshb-cre line.

Thyrotrope-specific Isl1 knockout mice are smaller than normal

Mice carrying cre recombinase under the control of the Tshb promoter, Tg(Tshb-cre), exhibit strong cre activity at e14.5 and excellent thyrotrope specificity (36). Mice with a floxed Isl1 allele (Isl1^loxP/loxP) were mated with Tg(Tshb-cre) mice, to produce Isl1^loxP/+; Tg(Tshb-cre) mice. We crossed Isl1^loxP/+;Tg(Tshb-cre) mice with Isl1^loxP/loxP mice to obtain Isl1^loxP/loxP;Tg(Tshb-cre) mice. Isl1^loxP/loxP;Tg(Tshb-cre) mice are mutant and Isl1^loxP/+ and Isl1^loxP/loxP were used as controls.

At P3, ISL1 colocalizes with more than half of the TSH immunopositive cells in the pituitary gland (white arrowhead) (Figure 3A). In the Isl1^loxP/loxP;Tg(Tshb-cre) mice there is little or no detectable coimmunostaining of ISL1 and TSH (n = 3). We quantified the number of cells with colocalized ISL1 and TSH out of the total number of TSH immunopositive cells. The 55% decrease in colocalization of ISL1⁺/TSH⁺ cells in the mutant mice is significant (P = .02). In controls, 58 ± 8% of the TSH cells coexpressed ISL1, but only 26 ± 11% did in mutants (Figure 3B). The efficiency of deleting ISL1 with Tshb-cre was less than observed for PITX2 or a cre reporter strain (36), which is not unusual (50). Nevertheless, it is clear that Isl1-deficient thyrotrope cells can survive and maintain Tshb expression.

Figure 3.

Isl1^loxP/loxP;Tg(Tshb-cre) mice present a significant growth insufficiency. A, Coronal pituitary sections from P3 Isl1^loxP/loxP;Tg(Tshb-cre) and control mice were immunostained with ISL1 (green nuclear stain) and TSH (red cytoplasmic stain). Cells with both ISL1 and TSH are present in the control sections (white arrowheads). There are very few cells with ISL1 and TSH in the Isl1^loxP/loxP;Tg(Tshb-cre) mice (white arrowheads). Photos were taken at ×630. Scale bar, 50 μm. B, The percentage of TSH cells that also express ISL1 (dICC, double immunopositive) was calculated for the Isl1^loxP/loxP;Tg(Tshb-cre) mutant and control mice. The mutant mice have significantly fewer colocalized ISL1⁺, TSH⁺ cells in their pituitaries compared with controls (*, P < .05). C, Average weights of male and female Isl1^loxP/loxP;Tg(Tshb-cre) and control mice at 5 and 8 weeks. There is a significant difference between both male and females at both ages. At 5 weeks, Isl1^loxP/loxP;Tg(Tshb-cre) males are 17.2 ± 3.2 g (n = 9), control males are 21.8 ± 1.0 g (n = 8), Isl1^loxP/loxP;Tg(Tsh-bcre) females are 13.5 ± 1.6 g (n = 10), and control females are 17.2 ± 1.5 g (n = 14). At 8 weeks, Isl1^loxP/loxP;Tg(Tshb-cre) males are 21.9 ± 2.6 g (n = 6), control males are 25.6 ± 1.4 g (n = 10), Isl1^loxP/loxP;Tg(Tshb-cre), and control females are 20.9 ± 1.5 g (n = 10). Statistical significance: ***, P < .001; **, P < .01.

We found that male and female Isl1^loxP/loxP;Tg(Tshb-cre) mice are significantly smaller than their normal littermates at 5 and 8 weeks of age (Figure 3C). At 5 weeks, the male and female Isl1^loxP/loxP;Tg(Tshb-cre) mice are 21% smaller when compared with control mice and at 8 weeks the male and female Isl1^loxP/loxP;Tg(Tshb-cre) mice are 14% smaller when compared with control mice. The weight differences at these times are statistically significant in females and males (P < .01, n = 6 or greater for each sex and genotype). This growth deficiency is consistent with expectations for moderate hypothyroidism. For both genotypes, males are larger than females, which is consistent with normal sexually dimorphic secretion patterns of GH.

Isl1^loxP/loxP;Tg(Tshb-cre) mice have reduced response to hypothyroidism

If ISL1 is necessary for thyrotrope differentiation and/or maintenance, then Isl1^loxP/loxP;Tg(Tshb-cre) pituitaries might have fewer thyrotropes than controls, or mutant thyrotropes might undergo hypertrophy in an effort to maintain homeostasis. To investigate these possibilities, we used immunohistochemistry to detect TSH in the adult Isl1^loxP/loxP;Tg(Tshb-cre) and control pituitary glands. There was no difference in the number or morphology of thyrotropes between the transgenic and control mice (Figure 4A).

Figure 4.

Blunted TSH response to hypothyroidism in Isl1^loxP/loxP;Tg(Tshb-cre) mice. A, There is no obvious difference in TSHβ staining in Isl1^loxP/loxP;Tg(Tshb-cre) pituitary glands at 8 weeks. TSHβ staining was performed on 5 pituitary sections at regular intervals for each transgenic and control mouse (n = 3/genotype). Photos were taken at ×100. Scale bar, 100 μm. B, qPCR was used to determine the mRNA expression fold change in liver Igf1 and pituitary Gh of Isl1^loxP/loxP;Tg(Tshb-cre) relative to controls. C, qPCR was used to determine the fold change in pituitary Tshb mRNA expression in Isl1^loxP/loxP;Tg(Tshb-cre) mice compared with controls in both basal and hypothyroid challenge conditions. Serum levels of TSH were measured in Isl1^loxP/loxP;Tg(Tshb-cre) and control females under basal and hypothyroid conditions. There was no significant difference in basal levels between Isl1^loxP/loxP;Tg(Tshb-cre) mice (297 ± 99 pg/mL, n = 6) and the controls (234 ± 114 pg/mL, n = 12), respectively. After hypothyroidism challenge, the transgenics responded with less dramatic elevation in TSH relative to controls: 27 × 10³ ± 13 × 10³ pg/mL, n = 4 vs 165 × 10³ ± 94 × 10³ pg/mL, n = 10. Note the logarithmic scale. **, P < .01. D, Total serum T₄ levels were measured in Isl1^loxP/loxP;Tg(Tshb-cre) and control male mice under basal and hypothyroid conditions. Under basal conditions, T₄ levels were significantly different (6.7 ± 1.3 ng/mL, n = 14, vs 5.1 ± 0.93 ng/mL, n = 7; *, P < .05). After hypothyroid challenge, the total T₄ levels in serum were reduced compared with basal levels (1.4 ± 0.57 ng/mL, n = 10, and 1.1 ± 0.29 ng/mL, n = 5).

To ensure that the growth defect is not due to disruption in the GH axis, we used qPCR to measure liver Igf1 expression and pituitary Gh expression of 8-week-old Isl1^loxP/loxP;Tg(Tshb-cre) and control mice. We saw no difference in expression of Igf1 or Gh (Figure 4B). We also used immunohistochemistry to examine GH protein in the mutant and control mice and saw no difference in the intensity or pattern of GH staining (Supplemental Figure 2).

The circulating levels of total T₄ provide one read-out of thyroid function. The Isl1^loxP/loxP;Tg(Tshb-cre) male mice have significantly lower basal levels of total T₄ compared with the control mice: 6.7 ± 1.3 ng/mL in controls (n = 14) vs 5.1 ± 0.93 ng/mL in transgenics (n = 7), P < .05, although the level is within the normal range (Figure 4D).

To assess the function of the thyrotropes, we measured serum levels of TSH and the amount of Tshb transcripts in the pituitary gland in female control and Isl1^loxP/loxP;Tg(Tshb-cre) mice at 8 weeks of age. We found no significant difference in the basal circulating levels of TSH (Figure 4C) or Tshb transcripts in the pituitary. We hypothesized that hypofunction of ISL1-deficient thyrotropes could cause the small body size of the mutants and low circulating T₄ levels. To test this idea we challenged the thyrotropes to determine whether they can respond to low thyroid hormone levels with elevated TSH. We induced hypothyroidism in female transgenic mice and controls (n = 6 transgenics vs 8 controls) by feeding them a low iodine diet enriched in a goiterogen, 0.15% PTU, for 4 weeks. To confirm that the treatment effectively induced hypothyroidism, we analyzed circulating total T₄ of transgenic and control mice treated with PTU. T₄ levels were dramatically reduced to less than 2 ng/mL (Figure 4D). We measured Tshb transcript levels and serum TSH levels in the PTU treated mice and found that although both transgenics and controls responded with elevated Tshb transcripts and serum TSH, the response in transgenics was 6-fold lower than in the controls: 27 × 10³ ± 13 × 10³ pg/mL in transgenics vs 165 × 10³ ± 94 × 10³ pg/mL in controls, P < .01 (Figure 4C). This demonstrates that Isl1^loxP/loxP;Tg(Tshb-cre) mutants exhibit a blunted response to hypothyroidism.

Isl1^loxP/loxP;Tg(Tshb-cre) mice have reduced thyroid function

Thyroid gland growth and function is directly affected by the action of thyrotropes. TSH stimulates the thyroid epithelial cells to produce thyroglobulin. The size of the colloid-filled follicle is a measure of thyroid activity. We examined H&E-stained sections of the thyroid gland in 8-week-old transgenic and control mice. The thyroid glands of Isl1^loxP/loxP;Tg(Tshb-cre) mice appear less developed than the control ones (n = 4, Isl1^loxP/loxP;Tg(Tshb-cre) and n = 3, Isl1^loxP/loxP) (Figure 5A). The overall size of the gland is smaller and the range of follicle sizes is smaller. To quantify the difference in thyroid gland morphology we measured the area of active thyroid follicles. The average follicular area of the Isl1^loxP/loxP;Tg(Tshb-cre) mice was 32% less than that of the control mice (P = .002) (Figure 5B). To ensure that the Tg(Tshb-cre) had no ectopic effect on the thyroid we examined ISL1 immunoreactivity in the thyroid glands. There was no difference in the number of ISL1⁺ cells in the Isl1^loxP/loxP;Tg(Tshb-cre) thyroid glands compared with controls (Figure 5C), suggesting that the reduced thyroid function is secondary to ISL1 deletion in the pituitary gland.

Figure 5.

Isl1^loxP/loxP;Tg(Tshb-cre) mice have impaired thyroid function. A, H&E stain of thyroid gland sections from 8-week-old Isl1^loxP/loxP;Tg(Tshb-cre) mice and controls. Photos were taken at ×100. Scale bar, 100 μm. B, The average follicular area in Isl1^loxP/loxP;Tg(Tshb-cre) was 221 ± 35 pixels (n = 4) vs control mice 325 ± 53 pixels (n = 3); **, P < .001 according to Mann-Whitney U test. C, ISL1 immunostaining (green) on thyroid gland sections from 8-week-old mice. Photos taken at ×630. Scale bar, 50 μm.

ISL1 is dispensable for gonadotrope maintenance and function

ISL1 is coexpressed with NR5A1 during development (Figure 2, E and F), but postnatally, ISL1 is not expressed in many gonadotropes (Supplemental Figure 3). To test the importance of ISL1 in gonadotrope function, we crossed Isl1^loxP/loxP mice with the previously described gonadotrope-specific cre strain: Tg(Lhb-cre) (35). The resulting Isl1^loxP/+;Tg(Lhb-cre) progeny were crossed with Isl1^loxP/loxP mice to obtain Isl1^loxP/loxP;Tg(Lhb-cre) mice with a conditional inactivation of Isl1 in differentiated gonadotropes. Isl1^loxP/loxP;Tg(Lhb-cre) male mice are the same size as normal males and larger than females, suggesting normal gonadal steroid effects on GH pulsatility. The Isl1^loxP/loxP;Tg(Lhb-cre) females have normal puberty as evaluated by vaginal openings, and both males and females have normal fertility (data not shown). This suggests that ISL1 is dispensable for maintaining mature, basal gonadotrope function.

Discussion

During development we detected coexpression of ISL1 with the gonadotrope lineage transcription factor NR5A1 but not the corticotrope marker TBX19. There was limited coexpression of ISL1 and POU1F1. This is consistent with reports of enrichment of ISL1 in gonadotrope-like cell lines, αT₃-1 and LβT₂, relative to corticotrope or somatolactotrope cell lines (33), and the reports on cell-specific expression in sheep and chick (31, 32). It is intriguing that ISL1 is required for development of pro-opiomelanocortin neurons, but it is not expressed in pro-opiomelanocortin cells of the pituitary (27). Colocalization of LH and NR5A1 with ISL1 was markedly reduced after birth. Despite the ability of ISL1 to act synergistically with NR5A1 and LHX3 to promote GnRH receptor expression in cell culture (33), we discovered that mice with Isl1 deleted in gonadotropes had normal puberty and fertility, suggesting that ISL1 is dispensable for mature gonadotrope function. ISL1 may have a role earlier in gonadotrope differentiation because it is coexpressed with NR5A1 at e14.5, and Lhb transcription is not detected until e17.5, which is the earliest time that Lhb-cre could act to delete ISL1. It is also possible that ISL1 has such a role in gonadotrope function that is compensated for by other LIM homeodomain transcription factors like LHX3, LHX4, or ISL2.

We present several lines of evidence that support a role for ISL1 in thyrotrope function. First, ISL1 expression colocalizes with most TSH cells in the postnatal pituitary. Secondly, Isl1 transcripts are increased in the Cga^−/− mice, which have extensive thyrotrope hypertrophy and hyperplasia (44). Third, deletion of ISL1 in thyrotropes using the Tg(Tshb-cre) line causes reduced body size, reduced circulating T₄ and reduced thyroid gland development. Moreover, these mutants exhibit a blunted thyrotrope response to hypothyroidism.

The efficiency of deleting Isl1 with Tshb-cre was less than observed for PITX2 or a cre reporter strain (36). This target site variability in cre efficiency is not unusual, and it can be confounded if there is a selective advantage for survival or proliferation of cells that escape the effects of the transgene (50, 51). Because residual ISL1⁺, TSH⁺ cells were present, ISL1 may be more important for thyrotrope function than suggested by the moderate growth insufficiency phenotype we observed.

The normal circulating TSH but low T₄ levels in Isl1^loxP/loxP, Tshb-cre mice is consistent with mild central hypothyroidism. The mutant pituitaries should have responded to the lower T₄ level with a rise in TSH, but they did not. This is consistent with the lower than expected rise in TSH in response to severe hypothyroidism. In the basal state, the slight challenge of modestly reduced T₄ levels did not elicit a rise in TSH, and although the challenge of severe hypothyroidism did provoke a response, the fact that it is 7-fold lower provides further support for the importance of ISL1 in pituitary-thyroid axis regulation. The phenotype of the Isl1^loxP/loxP, Tshb-cre mice is similar to the central hypothyroidism in TRHR and TRH mutant mice, which also present with low T₄ levels, relatively normal TSH levels, and impaired growth (52, 53).

POU1F1 is accepted as a critical transcription factor for specifying thyrotrope cell fate. Loss of function mutations result in severe TSH deficiency and the absence of TSH producing cells, except for a few transient TSH⁺, POU1F1⁻ cells that arise in the rostral tip of the developing gland (30, 54). POU1F1 activates Tshb gene expression together with GATA-binding protein 2 (55), and the spatial and temporal expression patterns of Pou1f1 and Tshb overlap. Pou1f1 transcripts first appear in cells located in the caudal medial region of the gland at e15.5, coincident with the appearance of Tshb transcripts. Thus, we were surprised that immunostaining revealed so many POU1F1⁻ thyrotropes in normal neonates. In contrast, most of the neonatal thyrotropes expressed ISL1, and some expressed NR5A1. Thus, there are clearly distinct, subpopulations of thyrotropes. In support of this, cells that express FSH, TSH, NR5A1, and LH receptor have been detected in pituitaries during late gestation and the neonatal period (56). These “thyro-gonadotrope” cells convert to functional gonadotropes in response to paracrine LH secretion, which triggers gonadotropin-releasing hormone receptor production and extinguishes TSH expression.

Pou1f1-cre mediated deletion of ISL1 did not have an obvious effect on growth or thyrotrope function. This could be attributable to less robust activity of Pou1f1-cre at e14.5–e16.5 compared with Tshb-cre (36). Another explanation is that there are many more ISL1⁺, TSH⁺ cells than ISL1⁺, POU1F1⁺ cells during early development and neonatal life. Perhaps the ISL1⁺, POU1F1⁻ thyrotropes are progenitors that progress to ISL1⁻, POU1F1⁺ thyrotropes. This would be consistent with the marked elevation in Isl1 expression that we observed in the pituitaries of mice with genetically engineered hypothyroidism, the Cga knockout mice.

Several transcription factors are necessary for thyrotrope function and normal growth. These include POU1F1, GATA-binding protein 2, PITX2, and ISL1 (1). The degree of growth insufficiency in both male and female Isl1^loxP/loxP;Tg(Tshb-cre) mice is modest compared with the severe dwarfism of Cga knockout mice, which do not secrete any bioactive TSH. However, the Isl1^loxP/loxP;Tg(Tshb-cre) mice have a more pronounced growth defect than we observed in the pituitary-specific knockout of Gata2 (57). Those Gata2-deficient mice had a modest growth defect in males but not in females, and the defect was only transient. The growth of Isl1^loxP/loxP;Tg(Tshb-cre) mice is also more impaired than that of mice with a conditional inactivation of Pitx2 in thyrotropes. The Pitx2^flox/−;Tg(Tshb-cre) mice have unchanged TSH and thyroid hormone levels in basal conditions, and a 2-fold reduction in TSH response after hypothyroidism challenge compared with controls (36). Taken together, these data indicate roles for multiple transcription factors in thyrotrope function, and they suggest a strong feedback mechanism to maintain normal homeostasis.

In the ganglion cells of the retina, ISL1 interacts synergistically with a POU homeodomain transcription factor, brain-specific homeobox transcription factor 3B, to promote cell differentiation (37). Because POU1F1 is the prototype POU homeodomain transcription factor expressed in the pituitary, and it is necessary for thyrotrope differentiation, we considered the possibility that ISL1 could interact with POU1F1 during pituitary development. Our preliminary results do not confirm this hypothesis, however. ISL1 did not interact synergistically with POU1F1 to increase transcription of either the Pou1f1 or Tshb promoters in HeLa cells (data not shown). There are many changes in transcription factor expression in Cga knockout mice, which, together with ISL1, could be involved in regulating the pituitary response to hypothyroidism (P. Gergics, unpublished data).

During development, several LIM domain transcription factors are necessary for a proper pituitary cell proliferation and differentiation, and elimination of one can affect the spatial and temporal regulation of the other factors. ISL1 and LHX3 expression overlaps early in pituitary development (e9.5–e10.5) but becomes mutually exclusive after that (29). Lhx3 null embryos have a transient lack of ISL1 expression at e12.5, a time that is critical for restricting ISL1 expression to the prospective anterior lobe (2, 6). However, ISL1 expression recovers in Lhx3 mutant pituitaries later in development (6). This suggests that LHX3 expression is necessary for ISL1 expression at e12.5. We hypothesize that delayed ISL1 expression contributes to the lack of thyrotropes in Lhx3 null mice.

In humans there are several known genetic causes of TSH deficiency, including mutations of genes coding for leptin receptor, TRH receptor, TSHβ subunit (58), IGSF1 (59) and multiple transcription factors (POU1F1, LHX3, LHX4, prophet of Pit1, and homeobox gene expressed in embryonic stem cells 1) (1). The ISL1 defect in mouse thyrotropes induces modest hypothyroidism and reduced growth. Therefore, mutations in Isl1 may contribute to pituitary thyroid axis dysfunction in humans. Individuals with homozygous loss of function mutations in ISL1 are expected to be nonviable due to the fact that ISL1 is involved in the development of several mouse organs (heart, pancreas, retina, motor neurons), and homozygous inactivation in mice is responsible for early death during embryogenesis. Hypomorphic mutations or those with a dominant negative effect might be compatible with life but be associated with pituitary based growth insufficiency. Patients with hypothyroidism and associated heart defects or diabetes may be most likely to have defects in ISL1. To date no systemic screening for ISL1 mutations has been carried out in patients with these features (60).

To conclude, our study shows that ISL1 is necessary for normal growth, pituitary thyrotrope response to hypothyroidism function, in addition to its previously known role in expansion of Rathke's pouch.